1992 Industry Outlook

High Technology

From Mainframes to Clones, a Tricky Time

1992 will bring turmoil and layoffs -- but a more competitive industry could emerge

John W. Verity in New York
Business Week

January 13, 1992

Perhaps the most popular characterization of the U. S. computer market in 1992 -- and easily the kindest -- is that it will be a year of transition. Suppliers at every level, from giant IBM to tiny PC clone makers, will be adjusting any way they can to the industry's rapidly changing economics. They will be shedding employees by the thousands, killing off marginal product lines, and teaming up to share costs. Industry revenues will grow about 6%, up from last year, but still short of the 10% or more in previous years, analysts say.

The painful but unavoidable fact is that an increasing majority of computers today are nothing more than small, cheap, high-volume commodities, composed of a mere handful of microchips and a core set of standard software, built using a tiny amount of labor. That's the main force that is halting revenue growth in traditional mainframes, driving profits to rock-bottom levels and forcing manufacturers to emphasize higher-margin software and services. Among the latter: help in designing and building networks and taking responsibility for all of a customer's data processing activities under a fixed-fee contract -- so-called outsourcing. Indeed, those two service sectors will be a bright spot this year, growing 20% or more, to about $ 18.8 billion in combined revenues.

Other factors, too, are holding back hardware sales. Among them are a lack of software standards for creating industrial-strength networks and reduced capital spending. Analysts say that Congress could help boost capital spending a little by bringing back the investment tax credit. In fact, analyst Marc Schulman at UBS Securities in New York knows customers who decided not to buy computers last year on the chance that the lawmakers will act this year.


In the absence of such a move, data processing budgets at large corporations will be up only 3.5% over 1991, vs. a projected 5.3% gain a year ago, according to a survey by CSC Index Inc., a Cambridge (Mass.) management consulting firm. The largest investments will be for systems to improve order processing, customer service, and product delivery.

In mainframes, revenues will probably be down some 3.3% this year, despite a major new-product cycle kicking in for mainframe leader IBM, according to market researcher Dataquest. IBM's large ES/9000 processors enjoyed fairly strong bookings through the fall and into early 1992. But orders look weak heading into the middle of the year, and heavy price discounting is rampant. One cause: Big Blue hasn't differentiated its new machines particularly well from earlier models, which often can supply the same computing power for half the price or less.

In midrange computers that sell for about $ 50,000 to $ 1 million, revenues won't be up more than 2% this year, says Dataquest. IBM's AS/400 and Digital Equipment Corp.'s VAX remain the two strongest sellers, but both face stiff competition from desktop machines.

Desktop sales will grow 25%, predicts Dataquest, with competition in that arena as fierce as ever. Workstation companies, such as Sun Microsystems, DEC, IBM, and Hewlett-Packard, will again play technological leapfrog as each launches machines based on the latest microprocessors. In PCs, prices will continue to fall, and suppliers will seek out new, high-volume distribution channels. The market will start to move to Intel's speed-of-sound 486 microchip, and everyone will be watching IBM's newly formed desktop computer unit. Overall, PC dollar sales may increase just 8.5%, compared with 12%-plus in recent years.

If one trend will distinguish 1992 from past years, it's the big move to client-server networks. These let desktop computers, or clients, share more powerful server computers that store masses of data, crunch numbers, and handle other specialized work. The economics of such setups are compelling, because the hardware's cost can be a tenth of that for a mainframe designed to do the same work. What's more, customers can expand such networks relatively easily, simply by adding extra computers as needed.


But what is still missing, for the most part, are programs to hold the networks together and to apply them to business problems. PCs, which are beginning to challenge workstations in hardware power, still lack the basic software -- or operating system -- needed to deliver full client-server functions. Customers using IBM-compatible PCs, for instance, are faced with a choice between installing Microsoft Corp.'s Windows 3.0 package now, waiting a few months to get IBM's OS/2 2.0 program, or waiting a bit longer for an all-new Microsoft package, Windows NT. They also can choose Sun's Solaris package, a version of AT&T's Unix operating system.

What customers want most, it seems, is to get machines of all brands, shapes, and sizes to interact effortlessly. And 1992 may see the start of just such a lovefest, as suppliers begin delivering software that adheres to a set of standards defining the so-called Distributed Computing Environment (DCE). Worked out by the Open Software Foundation, a consortium of suppliers including IBM, DEC, and HP, DCE defines the software plumbing needed for multisupplier networks. Also helping will be a new set of rules called Object Request Broker (ORB), endorsed by another industry consortium, that defines how different systems can share programs and complex data such as graphics and video. Together, DCE and ORB should help make client-server networks as easy to program as a single, self-contained computer.

From all the turmoil expected in the computer business this year, a leaner, more competitive, and potentially more profitable industry should emerge. If the economy picks up in the second half, computer makers could enjoy 10%-plus revenue growth in 1993 -- a first since 1988. That might make 1992 worth the struggle it promises to be. 


Copyright 1992 McGraw-Hill, Inc.